mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-15 17:46:45 +07:00
70fbe05745
I wanted to know the hottest path through a function and figured the branch-stack (LBR) information should be able to help out with that. The below uses the branch-stack to create basic blocks and generate statistics from them. from to branch_i * ----> * | | block v * ----> * from to branch_i+1 The blocks are broken down into non-overlapping ranges, while tracking if the start of each range is an entry point and/or the end of a range is a branch. Each block iterates all ranges it covers (while splitting where required to exactly match the block) and increments the 'coverage' count. For the range including the branch we increment the taken counter, as well as the pred counter if flags.predicted. Using these number we can find if an instruction: - had coverage; given by: br->coverage / br->sym->max_coverage This metric ensures each symbol has a 100% spot, which reflects the observation that each symbol must have a most covered/hottest block. - is a branch target: br->is_target && br->start == add - for targets, how much of a branch's coverages comes from it: target->entry / branch->coverage - is a branch: br->is_branch && br->end == addr - for branches, how often it was taken: br->taken / br->coverage after all, all execution that didn't take the branch would have incremented the coverage and continued onward to a later branch. - for branches, how often it was predicted: br->pred / br->taken The coverage percentage is used to color the address and asm sections; for low (<1%) coverage we use NORMAL (uncolored), indicating that these instructions are not 'important'. For high coverage (>75%) we color the address RED. For each branch, we add an asm comment after the instruction with information on how often it was taken and predicted. Output looks like (sans color, which does loose a lot of the information :/) $ perf record --branch-filter u,any -e cycles:p ./branches 27 $ perf annotate branches Percent | Source code & Disassembly of branches for cycles:pu (217 samples) --------------------------------------------------------------------------------- : branches(): 0.00 : 40057a: push %rbp 0.00 : 40057b: mov %rsp,%rbp 0.00 : 40057e: sub $0x20,%rsp 0.00 : 400582: mov %rdi,-0x18(%rbp) 0.00 : 400586: mov %rsi,-0x20(%rbp) 0.00 : 40058a: mov -0x18(%rbp),%rax 0.00 : 40058e: mov %rax,-0x10(%rbp) 0.00 : 400592: movq $0x0,-0x8(%rbp) 0.00 : 40059a: jmpq 400656 <branches+0xdc> 1.84 : 40059f: mov -0x10(%rbp),%rax # +100.00% 3.23 : 4005a3: and $0x1,%eax 1.84 : 4005a6: test %rax,%rax 0.00 : 4005a9: je 4005bf <branches+0x45> # -54.50% (p:42.00%) 0.46 : 4005ab: mov 0x200bbe(%rip),%rax # 601170 <acc> 12.90 : 4005b2: add $0x1,%rax 2.30 : 4005b6: mov %rax,0x200bb3(%rip) # 601170 <acc> 0.46 : 4005bd: jmp 4005d1 <branches+0x57> # -100.00% (p:100.00%) 0.92 : 4005bf: mov 0x200baa(%rip),%rax # 601170 <acc> # +49.54% 13.82 : 4005c6: sub $0x1,%rax 0.46 : 4005ca: mov %rax,0x200b9f(%rip) # 601170 <acc> 2.30 : 4005d1: mov -0x10(%rbp),%rax # +50.46% 0.46 : 4005d5: mov %rax,%rdi 0.46 : 4005d8: callq 400526 <lfsr> # -100.00% (p:100.00%) 0.00 : 4005dd: mov %rax,-0x10(%rbp) # +100.00% 0.92 : 4005e1: mov -0x18(%rbp),%rax 0.00 : 4005e5: and $0x1,%eax 0.00 : 4005e8: test %rax,%rax 0.00 : 4005eb: je 4005ff <branches+0x85> # -100.00% (p:100.00%) 0.00 : 4005ed: mov 0x200b7c(%rip),%rax # 601170 <acc> 0.00 : 4005f4: shr $0x2,%rax 0.00 : 4005f8: mov %rax,0x200b71(%rip) # 601170 <acc> 0.00 : 4005ff: mov -0x10(%rbp),%rax # +100.00% 7.37 : 400603: and $0x1,%eax 3.69 : 400606: test %rax,%rax 0.00 : 400609: jne 400612 <branches+0x98> # -59.25% (p:42.99%) 1.84 : 40060b: mov $0x1,%eax 14.29 : 400610: jmp 400617 <branches+0x9d> # -100.00% (p:100.00%) 1.38 : 400612: mov $0x0,%eax # +57.65% 10.14 : 400617: test %al,%al # +42.35% 0.00 : 400619: je 40062f <branches+0xb5> # -57.65% (p:100.00%) 0.46 : 40061b: mov 0x200b4e(%rip),%rax # 601170 <acc> 2.76 : 400622: sub $0x1,%rax 0.00 : 400626: mov %rax,0x200b43(%rip) # 601170 <acc> 0.46 : 40062d: jmp 400641 <branches+0xc7> # -100.00% (p:100.00%) 0.92 : 40062f: mov 0x200b3a(%rip),%rax # 601170 <acc> # +56.13% 2.30 : 400636: add $0x1,%rax 0.92 : 40063a: mov %rax,0x200b2f(%rip) # 601170 <acc> 0.92 : 400641: mov -0x10(%rbp),%rax # +43.87% 2.30 : 400645: mov %rax,%rdi 0.00 : 400648: callq 400526 <lfsr> # -100.00% (p:100.00%) 0.00 : 40064d: mov %rax,-0x10(%rbp) # +100.00% 1.84 : 400651: addq $0x1,-0x8(%rbp) 0.92 : 400656: mov -0x8(%rbp),%rax 5.07 : 40065a: cmp -0x20(%rbp),%rax 0.00 : 40065e: jb 40059f <branches+0x25> # -100.00% (p:100.00%) 0.00 : 400664: nop 0.00 : 400665: leaveq 0.00 : 400666: retq (Note: the --branch-filter u,any was used to avoid spurious target and branch points due to interrupts/faults, they show up as very small -/+ annotations on 'weird' locations) Committer note: Please take a look at: http://vger.kernel.org/~acme/perf/annotate_basic_blocks.png To see the colors. Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com> Cc: David Carrillo-Cisneros <davidcc@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@intel.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> [ Moved sym->max_coverage to 'struct annotate', aka symbol__annotate(sym) ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
329 lines
6.8 KiB
C
329 lines
6.8 KiB
C
#include "block-range.h"
|
|
#include "annotate.h"
|
|
|
|
struct {
|
|
struct rb_root root;
|
|
u64 blocks;
|
|
} block_ranges;
|
|
|
|
static void block_range__debug(void)
|
|
{
|
|
/*
|
|
* XXX still paranoid for now; see if we can make this depend on
|
|
* DEBUG=1 builds.
|
|
*/
|
|
#if 1
|
|
struct rb_node *rb;
|
|
u64 old = 0; /* NULL isn't executable */
|
|
|
|
for (rb = rb_first(&block_ranges.root); rb; rb = rb_next(rb)) {
|
|
struct block_range *entry = rb_entry(rb, struct block_range, node);
|
|
|
|
assert(old < entry->start);
|
|
assert(entry->start <= entry->end); /* single instruction block; jump to a jump */
|
|
|
|
old = entry->end;
|
|
}
|
|
#endif
|
|
}
|
|
|
|
struct block_range *block_range__find(u64 addr)
|
|
{
|
|
struct rb_node **p = &block_ranges.root.rb_node;
|
|
struct rb_node *parent = NULL;
|
|
struct block_range *entry;
|
|
|
|
while (*p != NULL) {
|
|
parent = *p;
|
|
entry = rb_entry(parent, struct block_range, node);
|
|
|
|
if (addr < entry->start)
|
|
p = &parent->rb_left;
|
|
else if (addr > entry->end)
|
|
p = &parent->rb_right;
|
|
else
|
|
return entry;
|
|
}
|
|
|
|
return NULL;
|
|
}
|
|
|
|
static inline void rb_link_left_of_node(struct rb_node *left, struct rb_node *node)
|
|
{
|
|
struct rb_node **p = &node->rb_left;
|
|
while (*p) {
|
|
node = *p;
|
|
p = &node->rb_right;
|
|
}
|
|
rb_link_node(left, node, p);
|
|
}
|
|
|
|
static inline void rb_link_right_of_node(struct rb_node *right, struct rb_node *node)
|
|
{
|
|
struct rb_node **p = &node->rb_right;
|
|
while (*p) {
|
|
node = *p;
|
|
p = &node->rb_left;
|
|
}
|
|
rb_link_node(right, node, p);
|
|
}
|
|
|
|
/**
|
|
* block_range__create
|
|
* @start: branch target starting this basic block
|
|
* @end: branch ending this basic block
|
|
*
|
|
* Create all the required block ranges to precisely span the given range.
|
|
*/
|
|
struct block_range_iter block_range__create(u64 start, u64 end)
|
|
{
|
|
struct rb_node **p = &block_ranges.root.rb_node;
|
|
struct rb_node *n, *parent = NULL;
|
|
struct block_range *next, *entry = NULL;
|
|
struct block_range_iter iter = { NULL, NULL };
|
|
|
|
while (*p != NULL) {
|
|
parent = *p;
|
|
entry = rb_entry(parent, struct block_range, node);
|
|
|
|
if (start < entry->start)
|
|
p = &parent->rb_left;
|
|
else if (start > entry->end)
|
|
p = &parent->rb_right;
|
|
else
|
|
break;
|
|
}
|
|
|
|
/*
|
|
* Didn't find anything.. there's a hole at @start, however @end might
|
|
* be inside/behind the next range.
|
|
*/
|
|
if (!*p) {
|
|
if (!entry) /* tree empty */
|
|
goto do_whole;
|
|
|
|
/*
|
|
* If the last node is before, advance one to find the next.
|
|
*/
|
|
n = parent;
|
|
if (entry->end < start) {
|
|
n = rb_next(n);
|
|
if (!n)
|
|
goto do_whole;
|
|
}
|
|
next = rb_entry(n, struct block_range, node);
|
|
|
|
if (next->start <= end) { /* add head: [start...][n->start...] */
|
|
struct block_range *head = malloc(sizeof(struct block_range));
|
|
if (!head)
|
|
return iter;
|
|
|
|
*head = (struct block_range){
|
|
.start = start,
|
|
.end = next->start - 1,
|
|
.is_target = 1,
|
|
.is_branch = 0,
|
|
};
|
|
|
|
rb_link_left_of_node(&head->node, &next->node);
|
|
rb_insert_color(&head->node, &block_ranges.root);
|
|
block_range__debug();
|
|
|
|
iter.start = head;
|
|
goto do_tail;
|
|
}
|
|
|
|
do_whole:
|
|
/*
|
|
* The whole [start..end] range is non-overlapping.
|
|
*/
|
|
entry = malloc(sizeof(struct block_range));
|
|
if (!entry)
|
|
return iter;
|
|
|
|
*entry = (struct block_range){
|
|
.start = start,
|
|
.end = end,
|
|
.is_target = 1,
|
|
.is_branch = 1,
|
|
};
|
|
|
|
rb_link_node(&entry->node, parent, p);
|
|
rb_insert_color(&entry->node, &block_ranges.root);
|
|
block_range__debug();
|
|
|
|
iter.start = entry;
|
|
iter.end = entry;
|
|
goto done;
|
|
}
|
|
|
|
/*
|
|
* We found a range that overlapped with ours, split if needed.
|
|
*/
|
|
if (entry->start < start) { /* split: [e->start...][start...] */
|
|
struct block_range *head = malloc(sizeof(struct block_range));
|
|
if (!head)
|
|
return iter;
|
|
|
|
*head = (struct block_range){
|
|
.start = entry->start,
|
|
.end = start - 1,
|
|
.is_target = entry->is_target,
|
|
.is_branch = 0,
|
|
|
|
.coverage = entry->coverage,
|
|
.entry = entry->entry,
|
|
};
|
|
|
|
entry->start = start;
|
|
entry->is_target = 1;
|
|
entry->entry = 0;
|
|
|
|
rb_link_left_of_node(&head->node, &entry->node);
|
|
rb_insert_color(&head->node, &block_ranges.root);
|
|
block_range__debug();
|
|
|
|
} else if (entry->start == start)
|
|
entry->is_target = 1;
|
|
|
|
iter.start = entry;
|
|
|
|
do_tail:
|
|
/*
|
|
* At this point we've got: @iter.start = [@start...] but @end can still be
|
|
* inside or beyond it.
|
|
*/
|
|
entry = iter.start;
|
|
for (;;) {
|
|
/*
|
|
* If @end is inside @entry, split.
|
|
*/
|
|
if (end < entry->end) { /* split: [...end][...e->end] */
|
|
struct block_range *tail = malloc(sizeof(struct block_range));
|
|
if (!tail)
|
|
return iter;
|
|
|
|
*tail = (struct block_range){
|
|
.start = end + 1,
|
|
.end = entry->end,
|
|
.is_target = 0,
|
|
.is_branch = entry->is_branch,
|
|
|
|
.coverage = entry->coverage,
|
|
.taken = entry->taken,
|
|
.pred = entry->pred,
|
|
};
|
|
|
|
entry->end = end;
|
|
entry->is_branch = 1;
|
|
entry->taken = 0;
|
|
entry->pred = 0;
|
|
|
|
rb_link_right_of_node(&tail->node, &entry->node);
|
|
rb_insert_color(&tail->node, &block_ranges.root);
|
|
block_range__debug();
|
|
|
|
iter.end = entry;
|
|
goto done;
|
|
}
|
|
|
|
/*
|
|
* If @end matches @entry, done
|
|
*/
|
|
if (end == entry->end) {
|
|
entry->is_branch = 1;
|
|
iter.end = entry;
|
|
goto done;
|
|
}
|
|
|
|
next = block_range__next(entry);
|
|
if (!next)
|
|
goto add_tail;
|
|
|
|
/*
|
|
* If @end is in beyond @entry but not inside @next, add tail.
|
|
*/
|
|
if (end < next->start) { /* add tail: [...e->end][...end] */
|
|
struct block_range *tail;
|
|
add_tail:
|
|
tail = malloc(sizeof(struct block_range));
|
|
if (!tail)
|
|
return iter;
|
|
|
|
*tail = (struct block_range){
|
|
.start = entry->end + 1,
|
|
.end = end,
|
|
.is_target = 0,
|
|
.is_branch = 1,
|
|
};
|
|
|
|
rb_link_right_of_node(&tail->node, &entry->node);
|
|
rb_insert_color(&tail->node, &block_ranges.root);
|
|
block_range__debug();
|
|
|
|
iter.end = tail;
|
|
goto done;
|
|
}
|
|
|
|
/*
|
|
* If there is a hole between @entry and @next, fill it.
|
|
*/
|
|
if (entry->end + 1 != next->start) {
|
|
struct block_range *hole = malloc(sizeof(struct block_range));
|
|
if (!hole)
|
|
return iter;
|
|
|
|
*hole = (struct block_range){
|
|
.start = entry->end + 1,
|
|
.end = next->start - 1,
|
|
.is_target = 0,
|
|
.is_branch = 0,
|
|
};
|
|
|
|
rb_link_left_of_node(&hole->node, &next->node);
|
|
rb_insert_color(&hole->node, &block_ranges.root);
|
|
block_range__debug();
|
|
}
|
|
|
|
entry = next;
|
|
}
|
|
|
|
done:
|
|
assert(iter.start->start == start && iter.start->is_target);
|
|
assert(iter.end->end == end && iter.end->is_branch);
|
|
|
|
block_ranges.blocks++;
|
|
|
|
return iter;
|
|
}
|
|
|
|
|
|
/*
|
|
* Compute coverage as:
|
|
*
|
|
* br->coverage / br->sym->max_coverage
|
|
*
|
|
* This ensures each symbol has a 100% spot, to reflect that each symbol has a
|
|
* most covered section.
|
|
*
|
|
* Returns [0-1] for coverage and -1 if we had no data what so ever or the
|
|
* symbol does not exist.
|
|
*/
|
|
double block_range__coverage(struct block_range *br)
|
|
{
|
|
struct symbol *sym;
|
|
|
|
if (!br) {
|
|
if (block_ranges.blocks)
|
|
return 0;
|
|
|
|
return -1;
|
|
}
|
|
|
|
sym = br->sym;
|
|
if (!sym)
|
|
return -1;
|
|
|
|
return (double)br->coverage / symbol__annotation(sym)->max_coverage;
|
|
}
|